Autonomous star sensing and pattern recognition for spacecraft attitude determination

Abstract

Autonomous star sensing and pattern recognition for attitude determination provides many technological challenges to modern spacecraft optical sensor design. This is mostly due to the relatively high accuracy requirements coupled with the faintness of many stellar sources, but is also due to real-time processing constraints. The performance of on-orbit star trackers is typically affected by nonlinearities such as lens distortion, coma, and chromatic aberration, as well as atmospheric refraction, thermal cycling, and possibly vibration. Despite these effects, the precise astrometric knowledge of inertially referenced stellar coordinates, along with thermoelectric cooling of the optical sensor, makes accurate star tracker calibration feasible. Land-based camera calibration, while not afflicted by many of the on-orbit difficulties, gives rise to a different set of problems relating to close range photogrammetry. The purpose of this dissertation is to report on the development and implementation of ideas related to near real-time, close range vision-based attitude sensing. The work begins with a survey of the current state of spacecraft attitude determination techniques, along with a discussion of relevant hardware devices. These ideas are extended to the case of close range photogrammetry for use in the laboratory, and a comprehensive discussion of the current experiments is presented. Topics include derivations of stellar and close range collinearity equations, mathematical modelling, CCD camera calibration techniques, resection and parameter estimation, optical aberrations, image processing and pattern recognition techniques, along with hardware and experimental results.